0

Another Gen6 or Generators API Mesh Error

I think we'd all agree that great software "feels" great because, from the users' perspective, it just works.  This happens when the developers take it upon themselves to play the role of magicians in so far as they create a facade that hides the complexities of the implementation from users. I think that Tinkercad users are generally well protected from implementation details. But not entirely. One area that I feel I and other users been unable to get the Tinkercad team to fully appreciate relate to the user-visible problems that crop up due to the complexities of the Gen6 kernel and architectural deficiencies in the shape generator API. Problems like these that have already been reported include these...

https://tinkercad.zendesk.com/hc/en-us/community/posts/204355559-Gen6-Clip-Bug 

https://tinkercad.zendesk.com/hc/en-us/community/posts/203839645-Mesh-Issues-with-Spheres-in-TinkerCAD 

https://tinkercad.zendesk.com/hc/en-us/community/posts/203814725-Tinkercad-Gen6-Mesh-Generation-Errors 

https://tinkercad.zendesk.com/hc/en-us/community/posts/202379529-Dimensions-and-rotation-of-generated-shapes-is-STILL-incorrect 

I appreciate the many challenges and risks inherent in changes to the kernel, any kernel. I also recognize the natural tendency of kernel-level engineers to wave off claims of kernel bugs (or compiler bugs) because of the high percentage of these reports that end up being user errors. On top of all that, characterizations of problems are often hard to come by and intermittent in nature. They're just hard to identify, hard to fix, and hard to regress.

With all this said, I have another similar report to make. I have a reliable characterization that involves a set of 3 shapes that are all instances of my MetaCapsule generator. The issue may well be a "rounding error" issue in either the kernel or in the API. Or it might be a error in "my" generator code. (But if the error is in my code, then we have a developer documentation issue, as this code was reviewed and corrected by Petr before it was published.)

The problem I'm experiencing, which is characterized in this drawing, is straightforward and easy to describe. All 3 shapes in the group are metaCapsules with the same Length, Width and Height properties and the same sized bounding box dimensions. Yet when grouped together as they are, the resulting mesh is faulty.

 

I view this error and the others I've mentioned above as significant and worthy of greater prioritization than they've seen because they expose tears in the wizard's curtain. The magic is broken as a result. 

  • Bob

2 comments

Please sign in to leave a comment.